Analysis was undertaken in SPSS (v18) using chi
squared and logistic regression to control for confounding
factors when examining associations between ecigarette
access and student demographics, smoking and
drinking behaviours. Hierarchical logistic regression was
used with school in the first block to control for clustering
of students within schools, with the exception of
analysis of regular smokers where sample sizes in individual
schools were insufficient. E-cigarette access was
identified through a ‘yes’ response to the question Have
you ever bought or tried electronic cigarettes? Smoking
status was established through a question asking students
which of a range of options best described them
(Table 1), with response options for smokers including
I only smoke when drinking alcohol and increasing levels
of regular smoking from 20 per day. Those smoking
=once a week, binge at any frequency). Students
were identified as users of specific drink types (e.g.
spirits) if they reported consuming at least one of that
drink type in a typical week.